AI wants to cure disease. It needs data first
QQQ•Biohub has brought together the US government, Google and Meta for a $1.8 billion effort to build a large cell dataset for AI research. The first dataset is expected in about a year, and usable models may be possible within five years.
1. Biology data gap
Researchers and scientists say AI-driven drug discovery is still in its early stages, with a lack of biological data a major obstacle. Current cell datasets contain hundreds of millions of cells, while an accurate predictive model may eventually need billions or trillions, Biohub science head Alex Rives said.
2. A $1.8 billion effort
Biohub is building a cell dataset with the US government, Google and Meta, bringing the project's total investment to $1.8 billion. Rives said the group aims to compress work that would normally take decades into five years, with a first dataset expected in about a year and usable models possible within five.
3. Testing AI predictions
Researchers will need to train models once the data exists and test whether they can accurately predict how cells respond to changes. Danaher plans to launch its first AI-powered autonomous research lab by early 2027, aiming to design, build and test antibodies up to eight times faster than conventional methods. The article also reports that biopharma companies spend roughly $140 billion annually on human clinical testing, and about 12% of drug candidates gain regulatory approval.
